Today's dataset looked at historical tornado tracks. Andy asked us to specifically focus on WHY within our exploration and analysis.
I started off by looking at the data dictionary and data. Everything in the data was abbreviated and so I took the time to read and understand the columns, re-name them and start thinking about what I'd like to focus on. As much as I've started with a plan on previous days, I have also struggled and then suddenly lost track of the idea of analysis and instead end up with lots of charts. I decided today that less is more and maybe I could answer less questions, focus on the why and have a decent dashboard to present (and a blog post!).
When - This for me always means a time series. When have these tornadoes occurred? What time do they occur the most? Are they seasonal? This was the starting point of my analysis. My heatmap showed some interesting trends and I became curious for context. Therefore, I thought that it would be interesting to add this context to support my analysis and lead on to my why.
Tornadoes form when warm, humid air collides with cold dry air. This got me thinking that weather must have some impact, and we keep hearing about rising temperatures. Has this led to tornadoes increasing? My why are tornadoes increasing looked at the number of tornadoes against North America land temperatures as well as Co2 output within the US.
Finally, I wanted to round this off with my last question, of who these tornadoes effect and how as well as what impact does this have. I kept this simple and thought to explore the overall fatalities and injuries across a map view whilst supporting this with the top 10 states. It's interesting that this supports my opening findings about the gulf stream and seasonality as the states affected fit right within it.
If I had more time, I think I would explore the remaining questions a lot further, possibly mapping the tornado paths and exploring other reasons why tornadoes may be occurring more frequently. I'd like more big numbers and analysis, however, when reading about the data and tornadoes, I learned that there are many constraints to some of these calculations as the way in which the data was collected changed over time.
Here's my final dashboard: